Overconfident CEOs and Capital Structure

Transcription

1 Master Thesis Financial Economics Overconfident CEOs and Capital Structure An empirical research on the US market Student name: Georgios Boutzias Student ID number: Faculty: Erasmus School of Economics Master s program: Economics & Business: Financial Economics Supervisor: Dr. Maurizio M. Montone Second assessor: Mr. Yuhao Zhu Date of final version: 3/7/2018

2 Abstract This study aims to investigate how does CEO overconfidence and other firm specific characteristics affect the decision making process regarding the capital structure of the firms. Trade-Off Theory, Pecking Order Theory and Market Timing Theory are the main capital structure theories that the analysis is based. The sample data includes US public listed firms and non identical CEOs. Panel Regression Analysis is implemented in order to derive the results. Two stock optionbased proxies of overconfidence have been constructed (Over and Over2). The results predict that the time invariant measure (Over) has a positive but insignificant effect on book leverage ratio, whereas the time variant measure (Over2) has a negative and significant effect on book leverage. Additionally, when I am testing for the prevalence of each capital structure theory, Pecking Order Theory seems to be more prevalent.

3 Acknowledgements I would like first to thank Dr. Maurizio Montone for his thorough contribution and his valuable guidance during the academic year of as my Master thesis supervisor as well as my lecturer of Behavioral Finance seminar. His instant, accurate and efficient remarks is something that helped me to achieve my goal and accomplish this Master thesis paper. I am really grateful! Moreover, I would like of course to thank my parents, Michael & Joanna, for their every day support and their advices. Next, I would also like to thank my best friends for their overall encouragement throughout the entire year as a Master student, and finally I also thank my fellow students that I met in this Master program.

4 Table of Contents 1. Introduction Literature Review Modigliani & Miller Theory Trade-Off Theory Pecking Order Theory Market Timing Theory Overconfidence Methodology Research Question, Hypothesis and Sub-Hypotheses Trade-Off Theory Sub-Hypotheses Pecking Order Theory Sub-Hypotheses Market Timing Theory Sub-Hypotheses Methodology Moneyness (Unexercised stock options) Testing for each Capital Structure Theory Testing for Trade-Off Theory Testing for Pecking Order Theory Testing for Market Timing Theory Data Sample Sample Reliability Summary Statistics Empirical Results Testing the main research question Testing for the sub-hypotheses of each capital structure theory Testing for the Trade-Off Theory Sub-hypotheses Testing for the Pecking-Order Theory Sub-hypotheses Testing for the Market Timing Theory Sub-Hypothesis Conclusion References... 55

5 1. Introduction Ever since, one of the most puzzling questions that firms have to answer about is which financing decisions do they have to follow in order to result on an optimal and beneficial capital structure. Capital structure in general is the amount of debt or equity that a firm has to employ in order to fund its operations and its investment decisions. Such operations could potentially be capital expenditures, mergers and acquisitions and other investments. In other words, firms have to answer the question of how much of the capital shall be debt and how much shall be equity so that they can run their business. Through the years, a lot of empirical research has been developed in order to address the question of what exactly determines firms financing decisions. These studies are trying to demonstrate which firm specific features can affect the capital structure of the firms and should be considered as important. Notably, some of the most important papers that try to address this puzzling problem are presented by Frank and Goyal (2009), Myers (1984), Titman and Wessels (1988) and Rajan and Zingales (1995). What they find is that firm specific features such as tangibility of the assets, profitability, liquidity, the size of the firm and growth opportunities do matter for its corporate structure. Initially, Modigliani and Miller (1958) tried to demonstrate with their paper that firm s total market value is irrelevant of its capital structure under some rigorous assumptions. Nevertheless, these assumptions are doubted due to the fact that markets nowadays they do have frictions. So, there was a need in the meanwhile for other theoretical frameworks to be developed in order to deal with such failures existing in the market and hence to help predicting which could be the optimal capital structure of the firms. Beginning with, Trade-Off Theory is a capital structure theory which argues that there is a tradeoff between tax shields and the cost of financial distress that a firm should balance when decides for debt financing. Next, Pecking Order Theory is related to information asymmetry. This theory argues that there is indeed a financial hierarchy that many companies follow. Briefly, firms tend to finance their structure using internal funds when they are adequate. If not, then these companies prefer first debt financing and then as a last financing resource the equity financing. Lastly, Market Timing Theory is a capital structure theory that has a more market view [1]

6 perspective. This theory tries to explain that managers often strive to time the market in order to make the right decisions to fund their corporate structure. In few words, the principal of this theory is that managers tend to issue equity when the stock price of their firm is relatively high and buy back equity when stock prices are relatively low. However, apart from firm specific determinants, also there might be other determinants that can influence the decision making process regarding the capital structure. Behavioral Finance can be the right field that can add to the existing literature of Corporate Finance. One common assumption of existing literature is that economic agents (e.g managers, CEOs, CROs, CFOs) are assumed to be fully rational (Fama, 1965). Obviously this assumption can be treated with skepticism and might turn to be problematic since it is widely observed that many financial distortions, either on a firm or country level, have been caused by irrational behavior choices. A lot of academic research has been focused on exploring why such managerial behavioral irrationality exists. Interestingly, many studies (Malmendier and Tate, 2005; Campbell et. al, 2011; Hirshleifer et.al, 2012) attribute this irrationality to the managerial behavioral bias of overconfidence. CEOs overconfidence is predicted to cause many investment distortions within the firm and this has an impact on its structure (Heaton,2002). Thus, it is interesting in this point to see how CEO overconfidence affects the capital structure decisions of the firm under the three capital structure theories stated above. This master thesis aims to add on existing literature by exploring and investigate the impact of being a CEO overconfident over the firm s capital structure decisions. Hence, I will examine how capital structure decisions can be affected if I incorporate the behavioral bias of overconfidence under the three capital structure theories argued briefly above. The working title of this thesis paper is Overconfident CEOs and Capital Structure and the main research question that I am trying to answer by conducting this empirical analysis is: How does CEOs overconfidence and other firm specific characteristics influence the decisions regarding the capital structure?. The reason that explains why I am motivated to select and conduct this research topic is because I find it really nice and challenging to derive and examine new findings by employing and combining insights from Corporate Finance literature, Behavioral Finance and Psychology. Since Behavioral Finance is a relatively new field of Economics, I would like to implement some aspects of this field onto [2]

7 conventional capital structure theories and observe how potentially irrational decisions can be explained. Additionally, I formulate a set of sub-hypotheses referred to each capital theory. In this way, I am trying to give a more in-detail analysis by testing which capital structure theory is more persistent if I bear in mind the effect of overconfidence proxies over book leverage. My sample consists of public listed US firms and non identical CEOs who are all operating in the US market. In the entire dataset, Financial- Insurance-Real Estate firms are not included based on Frank and Goyal (2009) and Baker and Wurgler (2002) papers. The sample period starts from 1992 till In order to obtain the data, I use the WRDS Compustat-IQ Capital database to get all the information that is needed for the capital structure of the firms and also the Compustat-Execucomp database from which I download all the necessary features regarding the construction of the option-holding based measures of CEO s overconfidence. Panel regression analysis has been used in order to conduct the analysis of this paper. The results of my empirical analysis predict that when I am running the main regression model using the first option-based measure of overconfidence, there is a positive relation between book leverage ratio and overconfidence despite the fact that the effect is not statistically significant. This relation is in line with existing literature and can be elaborated as that overconfident CEOs usually tend to underestimate the cost of financial distress, assuming that the firms they manage have less possibilities to bankrupt (Hackbarth,2009). This notion leads them to prefer more debt financing, seeking on the tax benefits they could gain with higher leverage ratios. By contrast, when I am applying the second measure of overconfidence, the results predict an inverse relation between book leverage ratio and overconfidence. The effect of this relation is statistically significant and can be explained considering some empirical behavioral findings as well. The negative relation can be attributed to the calculative nature of the second overconfidence proxy which is time-variant. Thus, it is more able to capture any changes on the behavioral traits through the years. A macroeconomic shock, like the World Financial Crisis of , could have had an impact on the behavior and the decisions of the CEOs. It is predicted that individuals that experience such financial crises, usually tend to be reluctant into participating on the markets. This change on the CEO s behavior, can lead to debt conservatism and selfsufficiency, avoiding debt financing. [3]

8 Additionally, testing for each capital structure sub-hypotheses, the results reveal that Pecking Order Theory seems to be a more prevalent capital structure theory, since the sensitivity of the core variables with the book leverage ratio, when firms are managed by overconfident CEOs, lead to the predicted empirical relations. The reminder of this paper is organized as follows: Section 2 demonstrate the current state of existing literature, Section 3 covers the methodological procedure I follow in this paper as well as the main research question, hypotheses and subhypotheses, Section 4 illustrates the sample description and descriptive statistics, Section 5 present the results of the empirical analysis and lastly in Section 6 the concluding remarks are presented. 2. Literature Review In this section, I am citing the most relevant theories referred to the capital structure as well as the recent empirical work of overconfidence. I will point out the most crucial key points of each theory and their main empirical predictions. One reasonable question that someone can ask is which exactly theory is more suitable and which one is more relevant with capital structure. Obviously, this is something that is debatable. Across the years many capital structure theories have been proposed, however only few of them have been supported thoroughly in academia. 2.1 Modigliani & Miller Theory Modigliani and Miller (1958) firstly introduced a theory regarding the capital structure in the late 1950s with their paper The Cost of Capital, Corporation Finance and the Theory of Investment. This pioneering theorem can also be expressed as the capital structure irrelevancy. They actually prove that capital structure of a firm is irrelevant of its value in an efficient market. Also the value of the firm is not influenced by the fraction of debt or equity that it holds in its books. In this paper they demonstrated 2 main propositions. The first one argues that The value of the asset does not change, regardless of how the net operating cash flows generated by the asset are distributed among different classes of investors (Modigliani & [4]

9 Miller,1958). In simple terms they argue that either you have the cash flows to pay the equity or the debt holders, in sense the cash flows generated in the firm do not change. The second proposition they state is that The cost of equity of a levered firm is equal to the cost of equity of unlevered firm plus a financial risk premium, which depends on the degree of financial leverage. All those propositions are made under some certain assumptions. In detail, Modigliani and Miller assume that i) Individuals borrow at the same rate as corporations do. ii) There is a perfect capital market. This means briefly that there is no information asymmetry and there are no transaction costs. iii) Additionally, they assume there are no any taxes implemented. iv) There is no cost of financial distress and lastly v) there is a fixed investment policy followed by each firm. These assumptions lead to the finding that the cash flows generated by the firm remain the same. Focusing on those assumptions made in their paper, it is instantly understood the reason why there are many critics about this theory. Almost all of them do not hold anymore and hence perfect capital market conditions, in which this theory is based, turn to be unrealistic. For example, it is widely accepted that nowadays information asymmetry between economic agents and transaction costs indeed exist. So, there is no perfect capital market as the Modigliani and Miller propose. Taxes have been imposed not only for individuals, but also for corporations. Lastly, one other inconsistency that derives from this theory is that it is not explicitly argued about the conflict that might appear between the agents and the shareholders. Usually, an agent acts on behalf of somebody else within the firm. They act on behalf of the shareholders of the firm. Agents (or managers) have as a target the maximization of the benefits of the shareholders where, in this particular case, those benefits come from the maximization of the cash flows generated by the firm. Shareholders, from their point of view, they want to make sure that managers that work on their firm have aligned incentives. Managers might have the tendency to invest more than what is optimal from the shareholder s perspective, and hence a conflict can exist. Modigliani and Miller in a later research paper, they indeed understand the need of reconsidering some strict and unrealistic assumptions, such as those represented above, and they argue in their paper that some world problems must be taken into account (Modigliani and Miller, 1963). In the meanwhile, later research also tried to deal with such market efficiency failures and tried to answer this puzzling question of the firm s optimal capital structure, incorporating other theories, which [5]

10 some of them will be explicitly presented right below. Optimal capital structure theories usually aim to find out which is the firm s optimal leverage over its assets that should hold so that it can take some benefits from that. 2.2 Trade-Off Theory After the Modigliani and Miller theorem, Trade-Off theory is concerned as the first solid capital structure theory. Myers (1984) in his paper assigns for the first time the name Static Trade Off Theory. One key difference with the Modigliani & Miller Theorem, however, is that the Trade Off Theory takes into account the fact that the there is a bankruptcy cost for each firm that might influence its capital structure decisions and their decisions of financing their investments and thus it tackles the Modigliani and Miller 1 st proposition. The core idea of this theory is that there is a tradeoff between tax-shields and the cost of financial distress so that the firm has to balance out and hence to have an optimal capital structure. In simple terms, this can be elaborated as the firm always selects its optimal capital structure considering that it can borrow up to the level that the interest tax shields and cost of financial distress are equal at the margin (Myers, 2003). The only reason that a firm might deviate from the optimal capital structure is the adjustment costs. Empirical evidences of the Trade-Off Theory shows that it can be tested crosssectionally using proxies for tax status and costs of financial distress (Myers, 2003). The question here is which proxies can be incorporated in the trade off theory. Fortunately, existing literature gives a comprehensive guideline regarding which proxies to select. Myers (2003), demonstrate in his handbook that proxies, such as business risk, non-debt tax shields and tangibility of the assets work reasonably well in cross sectional test of tradeoff theory. Taking a closer look on each of the proxies individually, we can see that using the business risk as a proxy of trade off theory is also been supported by De Jong et al. (2007) by stating in their paper that the business risk can count as a valuable feature that can affect the leverage of the firm. Usually, business risk can be attributed to the volatility of the earnings that a firm can generate. Non-debt tax shields, on the other hand, is the next proxy of trade of theory and its existence has also been supported in a recent work of Fama and French (2002). De Angelo and Masulis [6]

11 (1980), in their paper also show that the non-debt tax shields indeed affect the capital structure of a firm. Additionally, they state that non-debt tax shields can be used in order one firm to gain tax benefits by debt financing. Hence, they can be used as a good guide that helps into the financial investment decisions of the firm. Ultimately the last proxy that is incorporated into the Trade-Off Theory is the tangibility of the assets. Usually, tangible assets are used as collaterals for many companies in order to issue debt. Tangibility of the assets is also been used by Rajan and Zingales (1995) as one of the core variables of capital structure. What typically all these studies find about Trade-Off theory? In this point here, there is a need to bring out some of the most meaningful takeaways regarding the relation of each of the proxies of Trade-Off theory mentioned above with the leverage. Starting with, considering the business risk, we can understand that the more volatile are the earnings of the firm, the more risky the firm can be (it is also reasonable if we look this from the industry perspective). Thus, the business risk of the firm increases. However, if the business risk increases then the firm s cost of financial distress might increase as well. Titman and Wessels (1988), in their paper argue that the more volatile the earnings of the firm become, the less debt financing the firm chooses. In addition, according to Castanias (1983), the likelihood of a firm s bankruptcy might increase as far as the there is a (upwards) trigger of the earnings distribution (variance). Thus, he argues in his paper that there is an inverse cross sectional relationship between probability of bankruptcy and leverage (Castanias,1983). It make sense here to say that the firms, according to the empirical predictions mentioned above, should be reluctant into debt financing of their firm and thus they should keep the debt levels in a moderate proportion over its equity, otherwise it will be very difficult to fulfill all the obligations they have over their lenders. Hence, based on those empirical predictions I would expect as well a negative relation between business risk and leverage. However, another concern that I should take into account is that other studies like Ferri and Jones (1979), Titman and Wessel (1988),even though they initially expected the same negative correlation of business risk with the percentage of debt in a firm s financial structure, they found out that there is no significant relationship between them. Moreover, Scott (1976) in his paper finds that the correlation between leverage and earnings volatility can be sometimes ambiguous due to the different methods of calculation. [7]

12 The next proxy been used in the Trade-Off Theory are the non-debt tax shields. As it was stated again before, De Angelo and Masulis (1980) they argue that the firms in general can take the benefit of non-debt tax shields when they decide to finance their projects with debt. It is worth observing that many big size companies prefer to finance their investments or their finance structure with debt. This can be reasonably explained as they use debt financing since the interest they pay for the debt is tax deductable and so that they have tax benefits. The higher the interest payments, the higher the tax benefits the firms gain. Moreover, De Angelo and Masulis (1980) illustrate in their paper that apart from the tax benefits that one firm can gain from debt financing, non-debt tax shield can be used as an alternative way. Following the same pattern, if one firm has higher non-debt tax shields it is required then to pay lower taxes and apparently to have lower levels of debt. Therefore, I would expect a negative relation between the non-debt tax shields and leverage. Ultimately, the last proxy of the Trade-Off theory is the tangibility of the assets. As it was stated before, tangible assets are used as warrants from many companies in order to issue debt. Frank and Goyal (2009) mention that the tangible assets are an easy way for the lenders (Banks or Investors) to value the company. The more the tangible assets the firm has, the lower the distress cost in case of a bankruptcy will be. This is plausible since the lenders are able to pawn more assets from the firm they have lent their funds if the tangibility of those assets is high, and hence they can have a backup plan in case of the borrower goes bankrupt. Reasonably, I can make an argument that the more tangible assets the firm has, the more able this firm is to pursue more debt and thus, I expect a positive relation between firm leverage and tangibility of the assets. 2.3 Pecking Order Theory The next theory that tries to answer questions about how firms finance their investment decisions is the Pecking Order Theory. According to this theory, the firms in order to finance their projects and their investments, must choose between internal financing and external financing. Donaldson (1961) is assumed to be the inventor of the Pecking Order Theory and S. C. Meyers the one that started formatting this. This theory implies that firms, in order to finance their decisions must follow a specific [8]

13 hierarchy of funds. Myers (2003) shows that in general, firms prefer for internal financing rather than external. It is documented that most of the firms indeed follow this hierarchy especially these ones that are examined in the US market. Nevertheless, Meyers (1984) he states in his paper that there are some exemptions. By saying exemptions he means that there were some companies that led to finance their decisions by issuing equity, while at the same time it was easier for them to issue debt on an investment grade interest rate. The hierarchy of this theory demonstrates that if the internal funds are not adequate and external funds are needed for capital investment, firms will first prefer to issue debt before equity. As the demand for external financing is increasing, then the firm will be aligned with this hierarchy and will go through the Pecking Order Theory issuing from the safest to the riskiest form of debt, and as a final resource of fund, the firm will issue equity as the last resort (especially when the firm is under the pressure of financial distress). In order to demonstrate this in a more clear way, I will use some insights from Donaldson (1961), who was also a pioneer of the initial form of Pecking Order Theory. According to his research, Donaldson (1961) had as a sample U.S firms. In detail, he was conducting a survey in which he was asking managers of each firm about their financing decisions. The results show that there is indeed a financing hierarchy. Briefly, this hierarchy can be illustrated in the following way as: Firms tend to finance their decisions using, 1) Internal funds, 2)Debt, 3) Hybrid Securities and 4) Equity Issues. The reasons why such an hierarchy exists might be attributed to the transaction costs of raising capital and firm s debt capacity (Myers, 2003). This is contradictive to the Modigliani Miller Theorem that in the capital markets there are no any transaction costs. Moreover another consideration that Pecking Order Theory takes into account in contrast to Modigliani Miller Theorem is the fact that information asymmetry indeed exists and really matters for the capital structure. Additionally, Myers and Majluf (1984) made many assumptions in order to test for the Pecking Order Theory. Some of them are that i) managers have more information than outside investors, ii) transmitting information is costly, iii) management acts in interest of old shareholders iv) old shareholders are passive. Overall, the above arguments can be summarized by Shyam- Sunder and Myers (1999) finding that In its simplest form, the pecking order model of corporate financing says that when the internal funds of the firm are not enough to finance their decisions and its dividend commitments, then the firm should use debt [9]

14 financing, otherwise it can use equity finance in case only it issues junk bonds and the cost of financial distress at the moment is really high. Hence, in order to measure the Pecking Order Theory, I will use as a proxy again the tangibility of the assets since it can be used as an indicator of how difficult or not a firm is able to get debt financing from outside investors. Tangible assets, as it was discussed earlier, is a way for outside investors to approximate the value of the firm, and can use those tangible assets as collaterals in case that the firm goes bankrupt. Moreover, I will use the liquidity of the firm as an indicator of how much of the assets that the firm holds, are liquid or not. As the Pecking Order Theory predicts, corporations in general prefer first to use internal funds to finance their operations and then, if the internal funds are not enough, they will seek for external financing. Liquidity ratios can give a sneak preview of how liquid the assets of the firm are and therefore derive some extra information regarding the adequacy of the assets so that any potential firm can use them for internal financing. The last proxy that I will use in order to measure the Pecking Order Theory is the profitability of the firm. By using the profitability of the firm as a proxy, it is a way to show how much internal financing each firm has. According to Myers (2003) the more profitable firms tend to use less external financing not because their target leverage ratio is relatively low, but because in general those firms that are assumed to be more profitable, they usually have more internal funds available for financing their operations. In addition, those firms that are more profitable they indeed tend to minimize the amount of external financing they get over time (Frank and Goyal, 2009). The use of tangibility and profitability as proxies, have been supported also by Baker and Wurgler (2002). What the empirical studies of Pecking Order theory above predict about the relation of each of the three proxies with the leverage ratio of the firm? Starting with, the tangibility of the assets is used as an indicator that shows how much the firm can use its assets as collaterals when external finance is needed. The more tangible the assets are, the less the cost of debt issuance. However, due to information asymmetry, adverse selection exists and this leads firms to increase their debt capacity as far as the assets are more tangible. Hence I would expect in this case a positive relation between the tangibility of the assets and the leverage ratio of the firm. One remark here is that in Pecking Order Theory, the sign of tangibility might sometimes change to negative, due to the fact that firm with high tangibility may find equity issuance cheaper and thus substitute for equity rather than debt (Frank and Goyal, 2009). Next, [10]

15 regarding the liquidity ratio, as it was stated above the liquidity of the assets of the firm do matter a lot especially in the Pecking Order Theory. Let s suppose that one firm has high liquidity ratio. That implies that this firm holds in its balance sheet highly liquid assets such as cash, stocks, marketable securities, U.S treasuries, bonds and mutual funds. Remarkably, those types of assets are been characterized as liquid assets in sense that they can be converted to cash in a relatively short period. If this company is able to convert them into cash in a short period, simultaneously is also able to finance its structure or any other decision using its internal funds rather than using external finance like issuing debt (if and only if internal funds are adequate). Thus, I would expect a negative relation between the liquidity and leverage ratio. Lastly, the profitability is the remaining proxy of the Pecking Order Theory. Based on the empirical evidences mentioned above, like those ones from Myers (2003) and Frank and Goyal(2009) in which they state briefly that the more profitable firms tend to use more internal financing rather than external due to the adequacy of the assets, I would expect again a negative relation between profitability and leverage ratio. 2.4 Market Timing Theory Market timing theory, is a relatively simple theory of capital structure. Baker and Wurgler (2002) introduced with their paper the Market Timing Hypothesis and argued that this theory has many influences from other financial domains such as behavioral finance since it is a theory that predicts when it is exactly the right time for managers-companies to finance their firms structure in the most beneficial (costless) way. In addition, they argue that the ability to time the market actually gives some incentives to the managers to do it as efficient as they can if their incentives are aligned with those of the shareholders of the firm. In little words, capital structure includes all the aggregate of past attempts to time the equity market (Baker and Wurgler, 2002). Most empirical evidence shows that market timing really matters into the corporate financial policy of the firm, while at the same time all the other theories such as Trade-off Theory and Pecking Order Theory are still useful. Apparently, it does not seem as a complete theory of capital structure but it helps as a conditional theory. In contrast, Baker and Wurgler (2002) summarize in the end of their paper that the market timing hypothesis is the one that is dominating all the other [11]

16 theories. The main point of this theory is clear. Firms tend to issue equity when the stock price of their firm is relatively high and, vice versa, they repurchase equity when the stock price of their firm is relatively low. In detail, Baker and Wurgler (2002) state in their paper that it is observed that firms are more likely to issue equity when their market values are high, compared to book and previous market valuations, and they tend to repurchase equity when their market values are low. This can be expressed in another way as the firms tend to issue equity (stocks) when the cost of equity is low and repurchase (buy back stocks) when the cost of equity is high. Moreover they find that the effect of the market timing over the capital structure is very persistent across the years, and the current state of the firm s capital structure is strongly correlated to past historical market values. The main finding of their research is that those firms that have lower leverage ratio are actually these that issued equity when their market-to-book ratio was relatively high and, vice versa, highly levered firms are these that they got external financing when their market-to-book ratio was relatively low. As an additive supporting argument, Graham and Harvey (2001), in the survey which they were conducting, by asking 392 CFOs of companies in the US market, they find out that approximately 30% of the survey participants (CFOs) are taking seriously into account whether the stock of their firm is under/overvalued so that they can process into issue equity or not. Lastly, looking on the methodological procedure of Baker and Wurgler (2002) they notify that their regression model has as dependent variable the leverage ratio and as independent variable they use the marketto-book and a new variable market-to-book efwa, which stands for External Finance Weighted Average Market to Book. Actually, this new market-to-book variable is nothing more than a weighted version of the current market-to-book with a difference that when they say weighted, they indeed take into account the net equity issues of the firm and the net debt issues for each year for every firm. They state that the historical variations of the market-to-book can be summarized by market-to-book efwa. They incorporate simultaneously both the current market-to-book and the market-tobook efwa ( henceforth EFWAMB) because in this way they can observe the withinfirm time series variation and thus they document in this way the cumulative effect of the history of the market-to-book ratios on capital structure. What do empirical evidences predict for the market timing theory? According to the findings of Baker and Wurgler (2002), they show that those firms that raised the lowest external financing are those that have higher market-to-book ratio and vice [12]

17 versa, these firms that they have raised more external financing are those that have lower market-to-book values. Hence, I expect in my analysis that both book-to-market and EFWAMB will have a negative relation with the leverage ratio of the firm. 2.5 Overconfidence On the other hand, apart from the empirical Corporate Finance Theories such those presented explicitly above, it is really needed to take into account also some other aspects or theories that can have a meaningful influence into the decision making process specially when we look within a firm level. Thankfully, Behavioral Finance Theory is here to fill this gap. One assumption that is considered in the existing literature of Corporate Finance, is that agents (managers or investors) are assumed to be fully rational (Fama,1965). Obviously, this argument is not realistic and can be problematic. Behavioral Finance is a theoretical field that tries to explain anomalies in the financial markets and corporations based on psychology theorems. One well known anomaly (or bias) that is well presented and argued in the recent existing literature of Behavioral Finance is the Overconfidence bias. In general, overconfidence is a behavioral bias that leads individuals to overestimate their general knowledge on a certain field. Overconfident people also tend to underestimate their exposure over a certain decision or event and hence they underestimate the risk that emanates from this event. Also, they are very confident about themselves regarding their ability to control and manage various situations. These people can eventually be managers or CEOs in corporations operating in different industrial sectors and it is quite straightforward that this behavioral bias can have an impact on the investment decisions that their firm has to take. It is generally predicted that CEOs that have to deal with difficult tasks, such as making forecasts that are characterized by low predictability or multitasking roles within the firm level, tend to be highly overconfident. Overconfidence can be seen through two main channels, through miscalibration or through the better-than-average-effect. Ben- David, Graham and Harvey (2013) argue in their paper that miscalibration is the systematic underestimation of the range of potential outcomes and some potential reasons that allows miscalibration to exist as an behavioral bias are i) Managers overestimate their ability to predict the future and also ii) they underestimate the [13]

18 potential risk of an upcoming event which can be hidden. For example, miscalibrated investors underestimate the volatility of their firm future cash flows (Ben-David et al., 2013) The second channel that overconfidence can be observed is the better-thanaverage-effect. According to this, people in general think that their abilities or skills in a certain domain (business or society) are over performing the skills of other people. When individuals assess their relative skills, they tend to overstate their acumen relative to the average (Malmendier and Tate (2005); Larwood and Whittaker (1977); Svenson (1981)). CEO overconfidence can be linked also with corporate policies and corporate investments. As it is described in the Pecking Order theory, the funds needed for financing the corporate investments can be extracted either from internal financing (e.g cash flows generated) or external financing (debt issues / equity issues). Where can we see the linkage between CEO overconfidence and corporate financial policies? At this point it is worth to mention that such a behavioral bias like overconfidence can result in corporate investment distortions. By saying distortions I mean that a CEO can either underinvest or overinvest into some projects or investment opportunities. Underinvestment usually means that a CEO can withdraw from a project which in reality can have really good potentials in the long term as long as a good NPV value. On the other hand overinvestment is the case in which the CEO potentially can accept bad performing projects in reality, but phenomenally they seem to be valuable at the moment. Another reason that explains why such distortions exist is the information asymmetry between firm managers and outsiders (investors). Another view is that sometimes managers tend to overinvest in order to gain as much as more benefits creating the so called empire building. In this case, also due to info asymmetry, the targets of the shareholders with the targets of the managers are not aligned. One of the most meaningful relations that Ben-David et al.(2013) in their research survey find in the results, is that the more miscalibrated managers exist in the firm, the more debt financed is the firm they work on. Similarly, in the same notion also Malmendier and Tate (2005) with their outstanding paper find the same behavioral patterns. It is one of the first papers that examines and tries to understand in depth how CEO overconfidence can have a serious impact on corporate investment distortions. Heaton (2002) showed that common distortions in corporate investment decisions can be caused by managers who are overestimating the returns of their investments. Malmendier and Tate (2005) are doing so by testing their main hypothesis (H1: Investments of overconfident [14]

19 CEOs are more cash flow sensitive compared to those made by less overconfident CEOs ) and using different proxies that can measure the overconfidence of the CEO. Most notable proxy is the Holder67 overconfidence measure and then in rank is the Longholder in which the CEOs are classified as overconfident only if they keep their option holdings until the last year of its duration and lastly the Net Buyer proxy which classifies a CEO as overconfident if he was net buyer of company equity during the first five years they appear in the sample. One remark here is that both Longholder and Net Buyer are more difficult to be calculated due to the availability of the data. Some basic results that derived from Malmendier and Tate (2005) research paper are that there is a positive relation between CEO overconfidence and investment-cash flow sensitivity. Also another one notable finding is that the investment-cash flow sensitivity is more demonstrated in firms that are more equity dependent. On the other hand, overconfidence as a behavioral pattern has not only its cons but also some positive insights in terms of management. Hirshleifer et al. (2012) finds that overconfident CEOs are in general better innovators, since they can gain higher innovative performance for a given amount of R&D, and also they are able to translate growth opportunities into value creation for the firm, but with the restriction that these arguments hold true only if we talk about innovative industries that these CEO are incorporated in. Looking at all the information stated above as well as the empirical evidence regarding the overconfidence as a behavioral bias, I would expect a positive relation between overconfidence and leverage ratio. 3. Methodology 3.1 Research Question, Hypothesis and Sub-Hypotheses As it was stated before, the main research question which this analysis is seeking to answer is How does CEOs overconfidence and other firm specific characteristics influence the decisions regarding the capital structure?. At this point [15]

20 here, it is plausible to state the basic hypothesis, so that I can test my empirical results which will derive from the analysis in the following sections. H1: Firms that are managed by overconfident CEOs, will decide to finance their capital structure with more leverage compared to other corporations that are managed by less overconfident CEOs. However, based on the literature review presented in the previous section and the specific features induced from the theories referred to the capital structure, namely (A) Trade-off Theory- (B) Pecking Order Theory, (C) Market Timing I am also able to form some additional theory-specific sub-hypotheses. In this way I can check whether or not the empirical findings of the analysis are aligned with those hypotheses for each capital structure theory and their empirical predictions. Starting with, below they are presented some basic sub-hypothesis for the first theory of capital structure Trade-Off Theory Sub-Hypotheses H2: There is a negative relation between business risk and leverage ratio when firms are managed by more overconfident CEOs H3: There is a negative relation between non-debt tax shields and leverage ratio when firms are managed by more overconfident CEOs H4: There is a positive relation between tangible assets and leverage ratio when firms are managed by more overconfident CEOs To test for each sub-hypothesis mentioned above, I will apply the basic theoretical model based on the firm characteristics and overconfidence measurements over the entire data sample. One remark in this point here is that in order to test for them, I will apply the main regression model using each time interaction terms based on the overconfidence measures and the variable of interest for each capital structure theory. Hence, I will be able in this way to see in detail the sensitivity of those terms over the dependent variable. Those models will be explained in detail in Section (3.3). In the same way I can form also some basic sub-hypothesis, regarding the Pecking Order Theory and its specific features. [16]

21 3.1.2 Pecking Order Theory Sub-Hypotheses H5: There is a negative relation between the tangibility of the assets and the leverage ratio when firms are managed by more overconfident CEOs H6: The profitability of the firms and their leverage ratio are negatively related when they are managed by more overconfident CEOs H7: The liquidity of the firms and their leverage ratio are negative related when they are managed by more overconfident CEOs Lastly, I am forming the sub-hypotheses regarding the last theory of capital structure and its variable of interest that this theory incorporates Market Timing Theory Sub-Hypotheses H8: There is a negative relation between market-to-book and leverage ratio when firms are managed by more Overconfident CEOs. H9: There is a negative relation between EFWAMB and leverage ratio when firms are managed by more Overconfident CEOs. 3.2 Methodology In order to test the main hypothesis of the research question of this study and the additional sub-hypotheses, regarding each capital structure theory, it is worth mentioning in this point the main theoretical model that I will apply into my dataset. Considering the already existing literature in the Corporate Finance and Behavioral Finance field, I am able to select the core factors that I will incorporate in the main regression model. Such factors, regarding the capital structure, can be extracted from income statements and balance sheets of publicly listed firms in the US stock market. On the other hand, regarding the behavioral bias of overconfidence, I will construct an overconfident measure such as Holder 67 described in Malmendier and Tate(2005), plus another measure of overconfidence which is similar to Holder67 but with some different adjustments that have been applied. [17]

22 Due to the fact that my data set has a panel form, I assume that Panel Regression Analysis, using firm Fixed Effects and Year Fixed effects, is suitable to conduct this empirical research. One advantage of using a panel data model is that I can observe different dynamics in economic behavior. Thus, the main panel data model is presented below: LEV i,t = α i +α t +β 1 Size i,t-1 +β 2 Bus_Risk i,t-1 +β 3 ND_TaxShield i,t-1 +β 4 Tang i,t-1 +β 5 Profit i,t-1 +β 6 Liq i,t-1 +β 7 Over i,t-1 + β 8 M/B i,t-1 +β 9 EFWAMB i,t-1 +ε (1) Looking on Eq.(1), we can instantly understand that all the independent variables have 1-period lag compared to the dependent variable which is the book leverage ratio (LEV). The reason why there is such a case is that I want to examine and test the effect of each independent variable valued in the previous period over the dependent variable next period. In simple terms, I actually want to investigate the effect of the capital structure decisions taken and the behavioral bias (overconfidence) of each CEO in the previous period, over the leverage of the firm measured in the next period. Hence, it is crucial now to describe each variable of the main model, one by one, considering also which one refers to each of the 3 capital structure theories (Trade-Off theory, Pecking Order and Market Timing Theory). Starting with, the dependent variable is the leverage ratio, measured for the firm i and time t. Surprisingly, the leverage ratio is quite more puzzling than it seems to be. Thus, it is really important at this point to be defined. Frank and Goyal (2009) mention in their paper that there are several different ways that we can measure the leverage ratio. Specifically, they argue that differences between the leverage ratio measurements depend on whether the book or market values are taken into account. Moreover, they point out the differences between long-term debt and total debt. The Eq. (1) illustrated above can have either a market-based leverage or book based leverage. In recent literature, it is argued that there is no a rule of thumb for which one to select in order to be sure about the robustness of the outcomes. Frank and Goyal (2009) support this view as well. They state that there is no unified model of leverage currently available that can directly account for the six reliable factors of their analysis (Frank and Goyal,2009). [18]

23 In their analysis, Frank and Goyal (2009) select the market-based leverage in order to conduct their analysis. They support this view because only in this case they find reliable empirical patterns and a significant level for a set of six factors, which all account for more than 27% of the variation in leverage. However, there are also some other opinions regarding which kind of leverage is more suitable to select, and are contrary to the market-based leverage. Meyers (1977) indicates that the majority of the managers are actually more focused on book values and specifically in book leverage just because it is argued that the debt in general is better supported by assets in place rather than its own growth opportunities. Moreover, another claim that supports the selection of the book based leverage is that it is generally assumed that the markets are not always following a stabilized trend and most of the times the values that derive from market values are under scepticism and not reliable. Thus, even the market leverage can be considered as unreliable as well. Considering all the thoughts and concerns above regarding the leverage, for my analysis I will select the book based approach to measure the corporate leverage (LEV). In order to measure this ratio, I will divide the total debt by the total assets of the firm. By saying total debt, I actually imply that it equals to the current liabilities of the firm plus its no current liabilities. Focusing now on all the independent variables and coefficients, α i captures and controls for any unobservable effect, such as omitted firm characteristics or CEO characteristics that do not vary within a firm. Furthermore, α t captures the effect of aggregate time trends over the dependent variable. It actually controls for the effect of time-varying variables that are omitted from the model. Size: The variable Size describes the size of the firm. Apparently, also in this case there are several ways that can be used in order to measure the firm size. However, in my analysis, I will follow the approach of Frank and Goyal (2009), which measures the size of the firm as the natural logarithm of its total assets. This variable will be used as a control variable in the regression model. Under the Pecking Order Theory, size has an inverse relation with leverage according to Frank and Goyal (2009). Bus_Risk: This variable stands for the business risk that each firm faces due to its operations. One major risk for instance is the volatility of the earnings. According to many available ratios that can measure the business risk, the Financial Leverage [19]